🧠 All Memories

Created structured AI OS handoff package with canonical docs, project dossiers, task snapshot, JSONL memories, raw sources, and tar archive.

Created structured AI OS handoff package with canonical docs, project dossiers, task snapshot, JSONL memories, raw sources, and tar archive.

Analytics tracking was committed and pushed to origin/dev; Vercel/Plausible environment setup remains required.

Specialty-ranking routes verify locally but source table is empty.

Dermatology and surgery specialty-ranking routes verify locally, but cannot be marked done because med_school_specialty has zero rows.

First-party analytics and browser monitoring are dev-validated; server Sentry deferred because adapter-vercel tracing failed.

First-party analytics and browser monitoring are dev-validated; server Sentry deferred because adapter-vercel tracing failed.

MedSchools.ai has 26 users but no working visitor analytics; custom analytics_events table is empty and admin analytics has a likely schema bug.

Claude Code Discord setup now documented; channels require `/discord:access group add <channelId>` after DM pairing.

Project channels remain primary; Hermes owns normal interaction and Claude Code is invoked in-channel for dev/gstack work.

Private repo created for reproducible agent stack setup and MacBook-to-Mac Studio migration.

Claude plan strengthens Paperclip/governance and eval-loop specifics; Bob plan stronger on skepticism, project isolation, and vendor-neutral control plane. Updated recommendation combines both.

First-principles architecture for rebuilding the agentic startup OS: thin context, progressive memory, gstack workflow, graph-backed project intelligence, runtime competition.

Keep OpenClaw as gateway; upgrade Second Brain/gbrain-style memory; pilot Hermes; use Claude/Codex workers.

Recommendation to keep OpenClaw as always-on gateway, upgrade memory with gbrain-style synthesis, pilot Hermes as self-improving execution backend, and use Claude/Codex as worker runtimes.

# gstack Analysis & Integration Source: https://github.com/garrytan/gstack By: Garry Tan (YC CEO) Analyzed: April 1, 2026 ## Key Concept gstack turn

# MedSchools.ai Marketing Strategy ## Source Based on Greg Isenberg's "Stop Vibe Coding. Start Getting Customers" (March 2026) Video: https://youtu.b

# Autonomous Improvement Loops — Future Implementation ## Source Andrej Karpathy's `autoresearch` repo: https://github.com/karpathy/autoresearch Anal

# MedSchools.ai Lecture Notes — Product Spec (Idea Stage) ## Origin Henry Kim (CEO), March 11, 2026. Strategic expansion of MedSchools.ai from pre-me

# WiderWings Auto-Design Pipeline v1.1 **Owner:** Bob Wings **Date:** March 7, 2026 (updated from v1.0) **Status:** Active --- ## Overview A 5-ste

# Image Pipeline Setup — March 5, 2026 ## What End-to-end image generation pipeline for all WiderWings projects. Generate AI images → auto-resize for

# Agent Progressive Disclosure & Eval Framework — March 5, 2026 ## Changes Made ### 1. Progressive Disclosure (Anthropic pattern) All agent AGENTS.m

# Product Roadmap — WiderWings Inc **Last Updated:** 2026-02-11 **Owner:** Product Manager (currently Bob until PM agent is active) --- ## Vision

# Article Improvements & AI Feature Specs **Date:** 2026-02-15 **Source:** Henry feedback on "How Many Schools" article **Status:** Planning --- ##

# MedSchools.ai Scalability Analysis **Report Date:** February 11, 2026 **Analyst:** Sage Patel, Business Analyst **Project:** MedSchools.ai **

# MedSchools.ai Content Quality Audit ## Executive Summary **Date:** February 15, 2026 **Auditor:** AI Agent **Scope:** 9 published blog articles

# MedSchools.ai Database Utilization Audit **Audit Date:** February 11, 2026 **Conducted by:** Atlas Reeves (Backend Dev) **Database:** Supabase

# SEO Keyword & Competitive Analysis: Medical School Search Terms **MedSchools.ai** **Date:** February 15, 2026 **Analyst:** Subagent (SEO Researc

# MedSchools.ai SEO Content Strategy **Date:** February 12, 2026 **Prepared by:** Sage Patel, Business Analyst **Status:** Draft for Review ---

# Medical School Rankings Strategy - Executive Summary **Date:** February 11, 2026 **Analyst:** Sage Patel --- ## TL;DR: Build 5 Specialized Rank

# Medical School Rankings Strategy **Research Date:** February 11, 2026 **Conducted by:** Sage Patel (Business Analyst) **Status:** Draft for Revi

# WiderWings Recursive Development Protocol (RDP) v0.2 **Version:** 0.2 **Date:** March 4, 2026 **Author:** Bob Wings (Chief of Staff) **Based

# WiderWings Recursive Development Protocol (RDP) v0.2 **Version:** 0.2 **Date:** March 4, 2026 **Author:** Bob Wings (Chief of Staff) **Based

# MedSchools.ai — Complete System Documentation ## Overview MedSchools.ai is a personalized AI-powered platform helping pre-med students navigate the

# Hedge Trading Platform — Complete System Documentation ## Overview Hedge is a personal trading platform built for Henry Kim that automates portfoli

# WiderWings Recursive Development Protocol (RDP) *Draft v0.1 — March 3, 2026* *Author: Mark Wings (Research)* *Status: Draft for Bob's review* ---

# Recursive Software Development with OpenClaw ## What Is It? Recursive software development in OpenClaw means agents that build, test, improve, and

# PRD: Design Pipeline Tool (design.widerwings.com) **Status:** P2 — Parked for post-MedSchools.ai launch **Owner:** Bob **Created:** 2026-03-02 **Or

# AI-Powered Web Design Process *Strategy for WiderWings products (MedSchools.ai, CollegeDojo, Hedge, etc.)* Created: 2026-03-02 Source: Two YouTube

**Source:** https://www.youtube.com/watch?v=GDm_uH6VxPY **Tags:** ai-based web design --- Hi everyone, welcome to this agentic pattern [music] serie

**Source:** https://www.youtube.com/watch?v=86HM0RUWhCk **Tags:** ai-based web design --- Today I'm going to be showing you guys five simple hacks t

**Source:** https://youtu.be/oxob0x0Xz7s **Video:** The Ultimate Day Trading Guide (Full Training Chapters 1-10) --- [Music] in today's episode I'm

# How to Build Truly Expert AI Agents ## A Comprehensive Research Document for WiderWings *Research compiled Feb 27, 2026 by Mark Wings* *Sources: An

Research on AI agent best practices for Liz trading assistant — Top 10 improvements with priorities; key finding: extended thinking + reasoning scaffold are highest-impact quick wins

Comprehensive GEO strategy for MedSchools.ai. 4 phases: Foundation (llms.txt, robots.txt, FAQs, timestamps), Content Optimization (answer capsules, question headers, comparisons), API and Machine Access (public API, format=md, MCP server), Monitoring and Security. 8 tasks created on task board.

Covered calls, iron condors, VWAP mean reversion, and BB+RSI systems for range-bound markets. Henry specifically requested covered call deep-dive.

HEDGE PLATFORM INFRASTRUCTURE UPDATES (2026-02-22) DATA PROVIDER HIERARCHY (Henry-corrected): Schwab FIRST for ALL data (daily AND intraday) → Alpaca

OPENING POWER STRATEGY (Oliver Velez Method) Source: YouTube https://youtu.be/6XtBCqBhQ-k (transcribed via youtube-ultimate skill) Strategy doc: ~/cl

Henry corrected the data provider hierarchy. Schwab is PRIMARY for ALL data types (daily AND intraday), not just daily. Correct order: 1. Schwab (prim

Installed youtube-ultimate v4.2.2 from ClawHub. Uses youtube-transcript-api for free transcript extraction. Successfully transcribed 34KB Oliver Velez

Atlas: Fixed pre-signup onboarding persistence bug (3 root causes: device B data loss, OAuth data loss, MCAT format mismatch). 5 files fixed. Kai: Bui

Found 3 completed tasks never marked done: AI Assistant Panel (atlas, Feb 16), blog table styling (kai), 404/500 error pages (kai). All had completion

Daily: Schwab→yfinance. Intraday (1m/5m/15m/1h): Alpaca→yfinance. Unified provider at data_provider.py. Opening Power updated to use this. Henry confi

/strategy/opening-power page with 3 tabs: Narrow State Scanner, Backtester (configurable params, trade log, equity curve), Strategy Rules (visual Vele

Full strategy engine at backend/app/services/strategies/opening_power.py (750+ lines). State detector, bar classifier (elephant/tail/small), day simul

Transcribed Oliver Velez YouTube video (https://youtu.be/6XtBCqBhQ-k). Strategy: 2-minute bars with SMA20/SMA200 for NARROW (momentum) vs WIDE (contra

Replaced 1m/3m/5m chart intervals with single 2m chart. Fetches 3 days of 1m Schwab data and aggregates to 2m client-side (1 day expires on weekends).

Hedge Product Strategy approved 2026-02-21. Three Pillars: 1. Research (Charts, Screener, News, Pattern Scanner) - find opportunities 2. Strategies (

## Source - **URL:** https://youtu.be/pBS5vrqrUjk - **Channel:** Undiscovered Traders by Kinfo - **Guest:** Malik (@realTQQQtrader) — systematic/autom

## Goal 1. Transform Mark Wings into a world-class web scraper/researcher agent 2. Install and configure relevant skills from ClawHub and GitHub 3. Co

# Stock Trading Strategy Trends — February 2026 Research Report **Researcher:** Mark Wings 🦅 **Date:** February 18, 2026 **Sources:** Reddit (r/

Comprehensive research report on current stock trading strategy trends. Top strategies: (1) Wheel + LEAPS combo, (2) 0DTE SPX credit spreads, (3) mome

## Context Henry wants to turn me (Mark Wings) into a world-class scraper. Three layers: (1) Research & Recon — find where data lives, (2) Scraping &

## Strategy Ensemble System ### 10 Core Strategies 1. Dual MA Crossover (9/21) - Trend 2. RSI(2) Mean Reversion - Mean Reversion 3. MACD + RSI Combo

# Hedge Sprint 2: Strategy Ensemble System **Priority:** CRITICAL — Core value prop of Hedge **Owner:** Bob coordinating, Atlas (backend), Kai (front

## 4-Layer Architecture **Layer 1: Strategy Library** 10 strategies: MA Crossover, RSI(2), MACD+RSI, Bollinger, Momentum Rotation, ORB, ATR Turtle, V

## 10 Strategies to Implement | # | Strategy | Type | Timeframe | |---|----------|------|----------| | 1 | Dual MA Crossover (9/21) | Trend | Daily |

BUILT: HMM regime detection (5 regimes), regime API endpoints, strategy CRUD, 4 templates (Regime Momentum/Value/Volatility/Sector Rotation), backtest

Total: $664,199. Positions: VMFXX $337K (51% cash), VFIAX $79K, BLV $60K, VFISX $56K, VTSAX $31K, VENAX $29K, VIGAX $23K, VCSAX $18K, VMVAX $16K, VTMG

## Hedge Sprint 1 Backend Complete (2026-02-18) ### Database Changes (bthjgwuzkwqtodbxpufw) - Added `asset_class` column to `screener_stocks` (stock/

Screener uses FMP→Finnhub→Yahoo cascade with daily sync to Supabase

Cloudflare Tunnel config for Hedge backend with Quick Tunnel workaround

Alpaca dual account (Paper + Live) integration with separate credentials and services

Never use SQLite when Supabase is available - always use proper infrastructure

Python FastAPI chosen for Hedge backend due to superior ML/trading library ecosystem

Hedge trading platform fully set up with FastAPI backend, Supabase database, and Cloudflare tunnel

Comprehensive research on 10 algo/trading platforms for Hedge product direction. PLATFORMS ANALYZED: 1. QuantConnect - Code-based (Python/C#), LEAN

## PRD Additions (2026-02-16) ### 1. AI Integration **Models by Use Case:** - News Sentiment: Claude Haiku / GPT-4o-mini (fast, cheap) - Strategy An

## Broker Accounts | Account | Broker | API Access | |---------|--------|------------| | Trading | Alpaca | ✅ Yes (signed up 2026-02-16) | | Roth IRA

# Hedge: Algorithmic Trading Platform ## Product Requirements Document **Version:** 1.0 **Date:** 2026-02-16 **Owner:** Bob (Chief of Staff) **Stakeh

## Pattern Day Trader (PDT) Rules **Rule:** - Under $25K in margin account → Max 3 day trades per rolling 5 trading days - Over $25K → Unlimited day

## Notification Preferences **Daily:** - Summary of positions/performance - Recommendations for next day **Real-time:** - Trade alerts (when strateg

## Roth IRA ($92K) **Risk Appetite:** - Open to 3x leveraged ETFs (TQQQ, UPRO) - Open to momentum UP and DOWN (inverse ETFs like SQQQ) - Target: 20-3

## Account Structure | Account | Size | Access | Control | Target | |---------|------|--------|---------|--------| | **401K** | ~$668K | Vanguard onl

## Tools Evaluated ### Nano Banana (Google Gemini) - Gemini native image generation - Part of Gemini API (cheap at scale) - Good for: text-to-image,

## Use Cases ### 1. School Data Enrichment - Scrape official school pages: admissions, research, faculty, curriculum, scholarships, expenses, student

## Project Overview Programmatic trading platform for algorithmic investing. ## Brokerage Selection **Alpaca Markets** selected as broker: - SEC-regi

# HedgeFund: AI-Driven Stock Trading Platform Specification ## Executive Summary HedgeFund is a comprehensive AI-driven stock trading platform desig

# Inter-Gateway Agent Communication Setup Established: 2026-02-16 ## Architecture Multiple OpenClaw gateways coordinating via Tailscale private netw

# Bob Gateway Connection Info ## For Mark to call Bob directly: **Endpoint:** http://100.117.223.27:18789/tools/invoke **Headers:** - Authorization

# Mark Wings Node - Setup Complete (Feb 16, 2026) ## Machine - **Host:** mark-wings-t470s (ThinkPad T470s, Ubuntu Noble) - **User:** mark-wings ## T

# Agent Capability Setup Guide This document helps agents (especially Mark) become fully capable with WiderWings tools and services. --- ## 1. Seco

## Task 1: Updated How Many Medical Schools Article **Changes Made:** 1. Updated interview travel costs section to reflect virtual interview reality

# Top 5 MCAT Prep Resources for 2026 Research compiled Feb 2026 from provider sites and TestPrepInsight reviews. --- ## 1. Kaplan MCAT (Editor's Ch

Article improvements (tables, images, CTAs) + contextual AI assistant for PS feedback, Activities, School List strategy

Content audit reveals major gaps vs competitors: need 50-100% more content, 9 interactive tools, 50+ visuals. $30-45K investment = 150% traffic increase in 6mo.

Soft launch prep - Kai+Atlas deliverables done, SEO research complete, robots.txt fixed, Second Brain habits expanded

Top 30 med school keywords, competitive landscape (6 player types), 6 gaps to exploit, 3-phase content roadmap

Complete system architecture for MedSchools.ai covering tech stack, database schemas, caching, RAG, auth, billing

Complete Resend SMTP setup for MedSchools.ai auth emails. Sender: noreply@medschools.ai. Free tier (3K/month). Configured in Supabase Auth SMTP settings.

Stripe integration mostly complete. Pricing: Free/$49.95mo/$399yr + $99/mo interview. Remaining: coupons, paywalls, webhook completion.

Planning production launch milestones for MedSchools.ai. P0: auth fix, error pages, env vars. P1: analytics, monitoring. P2: SEO meta.

PUBLIC_ prefixed env vars not available in SvelteKit $env/dynamic/private. Use non-prefixed vars for server code.

Verified all Second Brain API endpoints working in production: auth, read, write, auto-embedding, semantic search

Generated and stored API keys for 5 agents (bob, kai, atlas, sage, maya) in .brain-keys file

Domain changed from memory.widerwings.com to brain.widerwings.com

Added project and agent filter dropdowns to all 7 Second Brain listing pages. Agent filter shows role from description field.

Changed from conservative to aggressive memory saving. Save too much, not too little. Cost of forgetting > cost of saving.

Decided on trigger-based memory saves (not periodic) to capture signal and avoid noise. Created BRAIN.md with explicit trigger rules.

Failed to save email auth work to Second Brain despite just building it. Lesson: save during work, not after.

Email auth fully implemented: login, signup, forgot/reset password, email confirmation. All routes and handlers built. Tested successfully with minor PKCE token timing issue.

Second Brain integration complete for all agents with API keys, trigger rules, and instructions

Decision to start with manual API inserts for conversation capture, then automate with cron job.